Hah would love to know how much of that is because of Lotus Field and NEH
I then also wanted to see if there was a difference between corp win rates before and after Upstalk. I’m not sure when Upstalk was released, but I think it was in early July.
You can figure out when each set was released on OCTGN by the version numbers
That’s helpful. It looks like the OCTGN “live” date was closer to June 8th, then.
I reran, and it didn’t change the results much. Both still significant, with a LSM Estimate of 0.2361 for all games and 0.3660 for Sigma1 games.
Ha! I just figured out why when I filtered for NEH games, the pack ID showed up as Future Proof. It’s because I match version strings by checking for x.x, rather than x.xx, and Future Proof is 3.1 whereas Upstalk is 3.11. So it thinks all of the Upstalk games happened in Future Proof.
I’d better fix that before I compute any winrates by pack rather than by ID!
Both players’ combined win rate of Runner and Corp or just the win rate of the side that’s being played?
Individual win rate was determined by a players performance over the entire dataset, regardless of side played. So if both players are sigma >= 1 players, then they both have a win percentage over 60% for all games played in the dataset.
Nice one, thanks. Good job 
Digging a little more. Now I’m looking at just “high skill” players, or those who are 1 standard deviation above the population mean (sigma>=1) and have games >=20 played. In addition, I’m only looking at games where both corp and runner had a Sigma >=1 player. Dataset is limited to the post-Upstalk era, June 8 and forward. 1632 total games.
Here’s a boxplot of the turns played during games. I’m not sure how OCTGN reports turns played, if it’s total turns or the number of turns for each side.
Next I wanted to see what identities were being used. Here is a side-by-side frequency of games played by each identity. I.E. Kate was used as the runner in 26% of all games played.
Next I wanted to see what identities were winning. Here is a side-by-side frequency of the win percentages for each ID out of a total of all faction wins. I.E. Andromeda accounts for 26% of all runner wins.
Lastly, here are the win percentages for Sigma >= 1 players by ID. ID’s were deleted if they had less than 20 games played. Totgames is the total number of games played (rounded because I’m lazy). I.E. Gabe won 61.8% of the 76 games Gabe played.
1 Like
Very interesting. Some thoughts:
Gabe’s high win percentage is interesting. Is this indicative of his power in the hands of skilled players? With only 5% of the games played, did he simply dodge poor matchups? He’s significantly above Andromeda, but without knowing which decks Andy and Gabe were matched up against, it’s hard to say more. Is there a way to break down which Corp IDs each runner was matched up against?
1 Like
It would be cumbersome to do it for everyone, but here’s a glimpse at Gabe specifically:
Of those 76 games where Gabe played, he was being piloted by 20 different people, with the highest being one person who won with him 9 times.
Here’s a side by side frequency of the Corp IDs. Games he won are on the left, games he lost are on the right.
Looking at this, he played a lot against ETF. So it’s certainly possible that Gabe is a good meta call right now, since ETF gets played a lot and he might just be very strong against them. It also shows that he is slightly better than 50/50 against NEH, which is pretty good compared to the rest of the field. Maybe because HQ is often lightly defended by NEH he can make use of his ability more? Just a guess.
It’s also possible that the results are due to random chance, though 76 games played should be enough. Maybe I’ll run that test next for practice (probably not today).
3 Likes
Fixed the version detection bug and pushed the changes up to GitHub, so now I’m ready to update my complete matchup plots. Probably not until after the conference I’m attending wraps up, though.
I can do some fun simple stuff, though. Corp / Runner aggregate winrates for each pack, all players:
Pack Games CorpWin RunWin
1 What Lies Ahead 2183 0.5144297 0.4855703
2 Trace Amount 8245 0.4812614 0.5187386
3 Cyber Exodus 10087 0.4990582 0.5009418
4 A Study in Static 17649 0.4882430 0.5117570
5 Humanity's Shadow 10031 0.4812083 0.5187917
6 Future Proof 27681 0.5055092 0.4944908
7 Creation and Control 11672 0.4674435 0.5325565
8 Opening Moves 42344 0.4948045 0.5051955
9 Second Thoughts 11551 0.4986581 0.5013419
10 Mala Tempora 8966 0.5070265 0.4929735
11 True Colors 21684 0.5059952 0.4940048
12 Fear and Loathing 27442 0.5261278 0.4738722
13 Double Time 25123 0.5393464 0.4606536
14 Honor and Profit 28905 0.5535720 0.4464280
15 Upstalk 44208 0.5690825 0.4309175
Looking at my updated competitive cut after fixing the version detection bug, the player count is different. I wasn’t expecting that to affect things, since I rate players on a weekly basis rather than a pack basis. So before I do matchup stuff I’ll need to double check that everything is working correctly now.
I’m also considering changing the Glicko deviation threshold I’ll accept, and maybe pruning the < 5 game players before doing ratings, which would definitely change the competitive cut results. I think it would be a good change, though. My intention is to cut the high shoulder off the deviation distribution and leave the flat portion, and I’m pretty sure that thresholding games will accomplish that. Then I’ll only be rating players who actually play.
Interesting breakdown. It’s nice to see how the swing towards corp wins happened over time. Hopefully we’ll see a swing back towards runner power with the rest of Lunar and into Order & Chaos like we see towards the end of Spin into H&P.
Thank you to everyone smarter than me who can process the data. Totally not surprised about Gabe being a boss. Imo it’s the toughest matchup for NEH, and can often beat long game corps before they can set up. I will however stubbornly keep playing Andy. 
We do love our consistency 
2 Likes
Thanks for indulging my curiosity!
Really? I have found (standard parasite or knight) Gabe to be upwards of 80% for RP Glacier against very strong players, (he has a much better matchup against HB Glacier). That’s about the only thing keeping me from switching to Gabe and looking for answers in shudder other factions.
That’s what I was saying. Gabe can often beat long game corps before they get a chance to set up their servers. If the Corp can setup however, Gabe is probably beat.
20% is a long ways off always ;D
1 Like
Haha, CAN was the operative word. Poorly written, I admit. *fixed for clarity